U.S. patent number 7,069,490 [Application Number 09/933,893] was granted by the patent office on 2006-06-27 for method for retransmission of lost packet in fading channels.
This patent grant is currently assigned to Fujitsu Limited, Tsinghua University. Invention is credited to Zhisheng Niu, Jing Zhu.
United States Patent |
7,069,490 |
Niu , et al. |
June 27, 2006 |
**Please see images for:
( Certificate of Correction ) ** |
Method for retransmission of lost packet in fading channels
Abstract
A method of retransmitting of a lost packet in a mobile
communication system in which multiple copies plus delays occur
i.e. after a transmitter has received from a receiver a negative
acknowledgement (NACK) signal, which points to a specific data
packet, multiple copies of the specific data packet are
retransmitted with a delay, which is inserted between two
consecutive copies. The number of multiple copies can be acquired
by a linear or exponential calculation according to the current
number of retransmission. The specific retransmission method may be
the interleaving transmission method, which transmits a
retransmission queue in the interleaving retransmission interval;
or the multiple queues polling transmission method, which transmits
the retransmission queue by copy queues sequence. In the correlated
fading channels, the method of the present invention can decrease
the number of retransmissions, and at the same time it can increase
efficiency of each retransmission.
Inventors: |
Niu; Zhisheng (Beijing,
CN), Zhu; Jing (Beijing, CN) |
Assignee: |
Tsinghua University (Beijing,
CN)
Fujitsu Limited (Kawasaki, JP)
|
Family
ID: |
4594115 |
Appl.
No.: |
09/933,893 |
Filed: |
August 21, 2001 |
Prior Publication Data
|
|
|
|
Document
Identifier |
Publication Date |
|
US 20020069388 A1 |
Jun 6, 2002 |
|
Foreign Application Priority Data
|
|
|
|
|
Oct 20, 2000 [CN] |
|
|
00130335 |
|
Current U.S.
Class: |
714/748; 370/216;
370/335; 370/394; 370/428 |
Current CPC
Class: |
H04L
1/0071 (20130101); H04L 1/1809 (20130101); H04L
1/1887 (20130101); H04L 1/189 (20130101) |
Current International
Class: |
H04L
1/18 (20060101); G08C 25/02 (20060101) |
Field of
Search: |
;714/748
;370/335,394,428,216 |
References Cited
[Referenced By]
U.S. Patent Documents
Other References
Bao, "Performance evaluation of TCP/RLP protocol stack over CDMA
wireless link," 1996, Kluwer Academic Publishers, Wireless Networks
vol. 2, Issue 3, pp. 229-237. cited by examiner .
Jeon et al., "Improved Selective Repeat ARQ Scheme for Mobile
Multimedia Communications," Feb. 2000, IEEE Communication Letters,
vol. 4, Issue 2, pp. 46-48. cited by examiner .
Weldon, "An Improved Selective-Repeat ARQ Strategy," Mar. 1982 IEEE
Transactions on Communications vol. COM-30, No. 3, pp. 480-485.
cited by examiner .
Jolfaei et al., "Improved Selective Repeat ARQ Schemes for Data
Communication," Jun. 1994, 1994 IEEE 44th Vehicular Technology
Conference, pp. 1407-1411. cited by examiner .
TIA/EIA/IS-707-A.2 (PN-4145.2), "Data Service Options for Spread
Spectrum Systems: Radio Link Protocol," Mar. 1999. cited by
examiner .
Fantacci, "Queuing Analysis of the Selective Repeat Automatic
Repeat Request Protcol Wireless Packet Networks," May 1996, IEEE
Transactions on vehicular technology, vol. 45. No. 2, pp. 258-264.
cited by examiner .
Konheim, "A Queueing Analysis of Two ARQ Protocols," Jul. 1980,
IEEE Transactions on communications, vol. com-28. No. 7, pp.
1004-1014. cited by examiner .
E. Dahlman, et al. "UMTS/IMT-2000 Based on Wideband CDMA", IEEE
Communications Magazine, Vo. 36, No. 9, pp. 70-80, Sep. 1998. cited
by other .
A. Chockalingam, et al. "Performance of TCP/RLP Protocol Stack on
Correlated Fading DS-CDMA Wireless Links" vol. 49, No. 1 IEEE
Transactions on Vehicular Technology, Jan. 2000, pp. 28-33. cited
by other .
W.S. Jeon, et al. "Improved Selective Repeat ARQ Scheme for Mobile
Multimedia Communications", IEEE Communications Letter, vol. 4, No.
2, Feb. 2000, pp. 46-48. cited by other .
Q.Q. Zhang, et al. "Finite-State Markov Model for Rayleigh Fading
Channels", IEEE Transaction Communications, vol. 47, No. 11, Nov.
1999, pp. 1688-1692. cited by other .
H.M. Chaskar, et al., "TCP Over Wireless with Link Level Error
Control: Analysis and Design Methodology", IEEE/ACM Transactions on
Networking, vol. 7, No. 5, Oct. 1999, pp. 605-615. cited by other
.
M. Zorzi, et al. "On the Accuracy of a First-Order Markov Model for
Data Transmission on Fading Channels" In proc. IEEE ICUPC'95,
Tokyo, Japan, pp. 211-215, Nov. 1995. cited by other.
|
Primary Examiner: Lamarre; Guy J.
Attorney, Agent or Firm: Katten Muchin Rosenman LLP
Claims
What is claimed is:
1. A method for retransmission of a lost packet in a fading channel
in a communications system including a transmitter and a receiver,
said method comprising the steps of: receiving at the transmitter a
negative acknowledgement (NACK) from the receiver, said NACK
pointing to a specific data packet; and, re-transmitting multiple
copies of the specific data packet with a delay inserted between
each consecutive copy wherein the inserted delay is a function of a
fading period.
2. The method according to claim 1, wherein with regard to said
multiple copies, the number of copies of the specific data packet
to be retransmitted is based on a number of times the specific data
packet has been re-requested (NACK-ed), wherein the number of
copies increases as the number of retransmissions increases.
3. The method according to claim 1, wherein the number of copies of
the specific data packet to be retransmitted increases linearly
along with the number of retransmissions, such that when the number
of retransmissions of the specific data packet is i.sup.th
retransmission, then the number of copies of the specific data
packet to be re-transmitted is i+1.
4. The method according to claim 1, wherein the number of copies of
the specific data packet to be retransmitted increases
exponentially along with the number of retransmissions, such that
when the number of retransmissions of the specific data packet is
i.sup.th retransmission, then the current number of copies of the
specific data packet to be re-transmitted is 2.sup.i.
5. The method according to claim 1, further includes the following
steps: setting at least two queues, including a transmission queue
and a retransmission queue, at the transmitter; storing new data
packets, which will be transmitted, in the transmission queue, and
storing copies of a data packet, which will be retransmitted, in
the retransmission queue; determining whether the retransmission
queue is empty or not; if the retransmission queue is empty, then
transmitting data packets in the transmission queue by a
first-in-first-out principle; and if the retransmission queue is
not empty, then transmitting the copies of the data packet in the
retransmission queue by interleaving transmission.
6. The method according to claim 5 wherein said transmitting the
copies of the data packet in the retransmission queue interleaving
transmission includes the steps of: setting a minimum value for a
length of time of an interleaving retransmission interval by a
timer; selecting from the retransmission queue one copy of each
retransmission data packet, and transmitting them in the
interleaving retransmission interval time length by the
first-in-first-out principle; and if within the minimum value of
the interleaving retransmission interval time length, all of the
selected one copy of each retransmission data packet has been
transmitted, then with the first-in-first-out principle,
transmitting the data packets in the transmission queue until the
interleaving retransmission interval is ended based on the minimum
value of the interleaving retransmission interval time length which
is set by the timer.
7. The method according to claim 6, wherein if within the minimum
value of the interleaving retransmission interval time length, all
of the selected one copy of each retransmission data packet in the
retransmission queue has been transmitted, and the transmission
queue is also empty, then transmission is stopped until the minimum
value, set by the timer, of the interleaving retransmission
interval time length ends this interleaving retransmission
interval.
8. The method according to claim 6, wherein when an interleaving
retransmission interval ends, if the retransmission queue is not
empty, then a new interleaving retransmission interval is started;
and if the retransmission queue is empty, then the interleaving
retransmission is ended and the data packets in the transmission
queue are transmitted using the first-in-first-out principle.
9. The method according to claim 6, wherein the length of the
interleaving retransmission interval is the larger of the following
two 1) said minimum value of interleaving retransmission interval
time length and 2) said maximum number of different data packets in
the retransmission queue.
10. The method according to claim 5, wherein said retransmission
queue includes copy queues, each copy queue holding one copy of
each specific data packet requiring re-transmission, the specific
data packets requiring re-transmission each having a sequence
number, and each copy queue holding the one copy of each specific
data packet requiring re-transmission according to the sequence
number.
11. The method according to claim 5, wherein said retransmission
queue includes copy queues, and said interleaving retransmission of
data packet copies in the retransmission queue further includes:
starting from a first copy queue of the retransmission queue,
transmitting by the first-in-first-out principle every data packet
copy in each copy queue until the last copy queue; and when all the
copy queues are empty, ending the interleaving transmission and
starting to transmit the data packets in said transmission
queue.
12. The method according to claim 11, wherein transmitting every
data packet copy in a next copy queue is executed only when the
current copy queue is empty, and then the next copy queue can be
transmitted.
13. The method according to claim 11, wherein the sequence of
transmitting from the first copy queue until the last copy queue,
includes copy queues which are not storing a data packet copy.
14. The method according to claim 11, wherein the number of copy
queues is equal to or greater than the number of retransmission
copies of the specific data packet.
15. A method for retransmission of a lost packet in a fading
channel, in a communications system including a transmitter and a
receiver, comprising the following steps: setting at least two
queues in the transmitter, including a transmission queue and a
retransmission queue; storing new data packets, which will be
transmitted, in the transmission queue, and storing the copies of
the lost data packet, the number of which is defined by a number of
retransmission, in the retransmission queue; determining whether
the retransmission queue is empty or not, and when the current
retransmission queue is empty, transmitting by a first-in-first-out
principle, data packets in the transmission queue; and when the
current retransmission queue is not empty transmitting the data
packet copies in the retransmission queue with an interleaving
transmission; setting a minimum value of time length of
interleaving retransmission interval by a timer, selecting one copy
of every retransmission data packet from the retransmission queue,
transmitting them in every interleaving retransmission interval by
the first-in-first-out principle; if before the end of an
interleaving retransmission interval, one copy of every
retransmission data packet in the retransmission queue has been
transmitted, then with first-in-first-out principle, transmitting
data packets in the transmission queue until the end of the minimum
value, set by the timer, of the interleaving retransmission
interval time length, and then, ending the interleaving
retransmission interval and starting next one interleaving
retransmission interval; if before the end of the minimum value of
the interleaving retransmission interval time length, one copy of
all data packets in the retransmission queue has been transmitted
and the transmission queue is empty, then stopping transmission
until the end of the minimum value, set by the timer, of the
interleaving retransmission interval time length, and then, the
interleaving retransmission interval will be ended and starting the
next interleaving transmission interval; and when an interleaving
retransmission interval is ended and the retransmission queue is
empty, then transmitting data packets in the transmission queue by
first-in-first-out principal.
16. A method for retransmission of lost packets in a fading
channel, in a communications system including a transmitter and a
receiver, the method comprising the following steps: setting at
least two queues in the transmitter, including a transmission queue
and a retransmission queue; storing new data packets, which will be
transmitted, in the transmission queue, and storing copies of lost
packets, the number of which is defined by a current number of
retransmission, in the retransmission queue; determining whether
the retransmission queue is empty or not, and when the
retransmission queue is empty, transmitting using a
first-in-first-out principle, data packets in the transmission
queue; and when the current retransmission queue is not empty,
transmitting the data packet copies in the retransmission queue by
multiple queue polling transmission; setting copy queues in the
retransmission queue with sequence number, each copy queue
including one copy of different data packet, starting from a first
copy queue, and with first-in-first-out principle, transmitting
every data packet copy in each copy queue in sequence, starting
only after sending out all the copies in one queue the next copy
queue transmission until the final copy queue, and when all the
copy queues are empty, ending the polling transmission and starting
the transmission of said transmission queue.
17. A method for retransmission of a lost packet in a fading
channel, in a mobile communication system including a transceiver,
comprising the steps of: transmitting by said transceiver a packet
to a receiver; when the transceiver receives information which
indicates that the receiver does not receive a specific packet
generating a plurality of copies of the specific packet; and
retransmitting by said transceiver the copies of the specific
packet in order at predetermined intervals the predetermined
intervals including an inserted delay wherein the inserted delay is
a function of a fading period.
Description
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention concerns an effective and reliable transmission
technique with a scheme for retransmission of a lost packet in
correlated fading channels.
2. Prior Art
At present, Internet and mobile communication technology has
development with a growing trend to converge them, thus requiring
mobile communication service, originally providing voice
transmission service only, to also provide data transmission
service. Because of correlated fading characteristic of a wireless
channel, data packets can be lost, so there is a problem of
unreliable transmission. Reliable link layer protocols, such as
Automatic Repeat Request (ARQ), are one method to provide reliable
transmission in an unreliable transmission system. ARQ methods can
be roughly classified into Stop-and-Wait (SW), Go-back-N (GN) and
Selective Repeat (SR). Among them, SR-ARQ is the most efficient and
has been widely used in practical mobile systems, for example IS-99
(TIA/EIA/IS-99, "Data Services Option Standard for Wideband Spread
Spectrum Digital Cellular System", 1955).
FIG. 1 shows a working procedure of Selective Repeat (SR-ARQ). When
the receiver sends back the Acknowledge (ACK) or Negative
Acknowledge (NACK), the transmitter determines whether a specific
data packet (for example #1) is lost or not, and makes a selective
retransmission. When the transmitter receives an ACK signal, it
means the #1 packet has been received successfully, it is
unnecessary to retransmit again. When the Transmitter receives a
NACK signal, it means that the #1 packet has not been received
successfully, it is a transmission failure, the packet is lost and
it is necessary to retransmit. When there is retransmission, each
time only one copy of the lost #1 packet is retransmitted. The
figure shows that after two times retransmission of #1 packet copy,
it is received successfully. This means that in the traditional
SR-ARQ scheme, when a packet is lost, only one copy of the lost
packet is retransmitted each time. Obviously, the worse the
environment of transmission, the more times retransmission is
needed. In this case the data packet has a longer persistence time
in the transmitter buffer and will seriously decrease the quality
of data service.
Naturally, when a data packet is lost, multiple copies of the lost
data packet can be retransmitted each time. But in a mobile
communication system, because of the instinctive burst error
characteristics in correlated fading channels data packets are
successively lost and if multiple copies are sent for each
retransmission they will meet the same bad state of the fading
channels at the same time and the retransmission is failure, i.e.
retransmission efficiency is low. Therefore, in correlated fading
channels the issue of decreasing the number of retransmissions and
increasing retransmission efficiency are two big problems needing
consideration.
SUMMARY OF THE INVENTION
It is an object of the invention to provide a retransmission method
for lost packets in correlated fading channels that will decrease
the number of retransmissions and increase the efficiency of each
retransmission.
The above and other objects of the invention are implemented as
follows. A retransmission method for lost packet in a fading
channel is characterized in that: when the transmitter receives
from a receiver in a mobile communications system, a negative
acknowledgement (NACK), which points to a specific data packet,
multiple copies of the specific data packet will be retransmitted,
wherein a delay is inserted between two consecutive copies.
The number of multiple copies is acquired by calculation based on
the number of current retransmissions of the specific data packet.
The number of copies is increased along with an increase of the
number of retransmissions.
In one embodiment, the method of the present invention used to
retransmit lost packets in a fading channel, includes the following
steps: A. At least two queues are set in the transmitter, including
a transmission queue and a retransmission queue; B. New data
packets, which will be transmitted, are stored in the transmission
queue; the copies, the number of which are defined by current
number of retransmission data packets, are stored in the
retransmission queue; C. It is determined whether the
retransmission queue is in the state of empty or not, when the
current retransmission queue is empty, with first-in-first-out
principle, data packets in the transmission queue are transmitted;
When the current retransmission queue is not empty, the data packet
copies in the retransmission queue are transmitted with
interleaving transmission; D. The minimum value of time length of
an interleaving retransmission interval is set by a timer, select
one copy of every retransmission data packet from the
retransmission queue, transmit them in every interleaving
retransmission interval by the first-in-first-out principle; if
before the end of a minimum value of interleaving retransmission
interval time length, one copy of all data packets in the
retransmission queue has been transmitted, then with the
first-in-first-out principle, data packets in the transmission
queue are transmitted until the end of the minimum value, set by
the timer, of the interleaving retransmission interval time length.
Then this interleaving retransmission interval will be ended and
the next one will be started; if before the end of the minimum
value of the interleaving retransmission interval time length, one
copy of all data packets in the retransmission queue has been
transmitted and the transmission queue is empty, then transmission
stops until the end of the minimum value, set by the timer, of the
interleaving retransmission interval time length. Then, this
interleaving retransmission interval will be ended and the next one
will be started; when an interleaving retransmission interval is
ended and the retransmission queue is empty, then data packets in
the transmission queue are transmitted by the first-in-first-out
principle.
Another embodiment to retransmit a lost packet in a fading channel
includes the following steps: A. At least two queues are set in the
transmitter, including a transmission queue and a retransmission
queue; B. New data packets, which will be transmitted, are stored
in the transmission queue; copies, the number of which is defined
by current number of retransmission data packet, are stored in the
retransmission queue; C. It is determined whether the current
retransmission queue is in the state of empty or not, when the
current retransmission queue is empty, with the first-in-first-out
principle, data packets in the transmission queue are transmitted;
when the current retransmission queue is not empty, the data packet
copies in the retransmission queue are transmitted with multiple
queue polling transmission; D. Setting copy queues with a sequence
number, each copy queue includes one copy of different data packet.
Starting from the first copy queue, with the first-in- first-out
principle, every data packet copy in each copy queue is transmitted
in sequence.
Only after sending out all the copies in one queue, the next copy
queue transmission can be started until the final copy queue. When
all the copy queues are empty, then the polling transmission is
ended and the transmission of the said transmission queue is
started.
According to yet another embodiment a transceiver, in a mobile
communication system, transmits a packet to a receiver or provides
a plurality of copies of a special packet, when the transceiver
receives information which indicates that the receiver does not
receive the specific packet, the transceiver retransmits the
specific packet in order at a predetermined interval.
In another embodiment, the method of the invention used to
retransmit a lost packet in a fading channel, is a retransmission
method with multiple copies plus delay. By transmitting multiple
copies of a lost data packet for each retransmission, it is
different than the traditional SR-ARQ, in which only one copy is
sent, therefore the success probability of each retransmission is
increased. At the same time, an adequate delay is inserted between
two consecutive copies of the same lost data packet when
transmitting multiple copies of a specific lost data packet. In
this way the probability of meeting the bad state of the fading
channel for multiple copies of the one lost packet, is decreased,
i.e. success probability of each retransmission is increased.
Therefore efficiency of retransmission is increase effectively and
number of retransmission is decrease.
Following combines embodiment and appended figures to further
describe technology of the invention.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a schematic diagram of a typical SR-ARQ working
procedure;
FIG. 2 is a diagram to define the number of copies for each
retransmission by a linear increasing scheme according to the
invention;
FIG. 3 is a diagram to define the number of copies for each
retransmission by an exponential increasing scheme according to the
invention;
FIG. 4 is a schematic diagram of an interleaving procedure within
an interleaving retransmission scheme according to the
invention;
FIG. 5 is a diagram representing an implementation scheme for an
interleaving retransmission scheme according to the invention;
FIG. 6 is a diagram, which is another implementation scheme for an
interleaving retransmission scheme according to the invention;
FIG. 7 is an analysis diagram of effectiveness for an interleaving
retransmission scheme according to the invention;
FIGS. 8 10 show examples of the emulated relationship curve
diagrams between Delay Time with Effective Throughput, Delay Time
with a Mean of the Number of Retransmissions and Delay Time with
Variation of the Number of Retransmissions, respectively; and
FIG. 11 shows an example of the emulated curve diagram of the
improvement of transmission performance after using the invention
scheme.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
FIG. 1 has been mentioned above, it would not be repeated.
Referring to FIG. 2 and FIG. 3, these figures show two schemes of
the invention respectively, which define the specific number of
lost data copies for each retransmission. They include a linear
increase scheme, as shown in FIG. 2, and an exponential increase
scheme, as shown in FIG. 3, they are all related to the number of
retransmissions of this (current) time. The linear increase scheme
can be summarized as the number of copies for i.sup.th
retransmission is i+1. The exponential increase scheme can be
summarized as the number of copies for i.sup.th retransmission is
2.sup.i.
In FIG. 2, when a transmitter receives a NACK signal the first time
for # number data packet, it retransmits two copies of # number
data packet, and inserts a delay d between two consecutive copies
of each # number data packet. When the transmitter receives a NACK
signal the second time for # number data packet, it retransmits
three copies of # number data packet, and inserts a delay d between
two consecutive copies of each # number data packet; and so on. If
the transmitter receives a NACK signal the third time for # number
data packet, it will retransmit four copies of # number data
packet, and will insert a delay d between two consecutive copies of
each # number data packet (this has not been shown in the figure).
FIG. 2 shows at the second time of retransmission, its third copy
is received correctly.
In FIG. 3, when a transmitter receives a NACK signal the first time
for # number data packet, it retransmits two copies of # number
data packet, and inserts a delay d between two consecutive copies
of each # number data packet. When the transmitter receives a NACK
signal the second time for # number data packet, it retransmits
four copies of # number data packet, and inserts a delay d between
two consecutive copies of each # number data packet; and so on. If
the transmitter receives a NACK signal the third time for # number
data packet, it will retransmit eight copies of # number data
packet, and will insert a delay d between two consecutive copies of
each # number data packet (this has not been shown in the figure).
FIG. 3 shows at the second time of retransmission, its third copy
is received correctly.
Obviously, for the exponential increase scheme the number of copies
is increased rapidly as the retransmission numbers increase,
therefore efficiency is lower, but there is an improvement of a
decrease of retransmission numbers and persistence time. This
scheme is more appropriate in an environment where the channel
condition is very bad and the propagation time is longer.
The delay d, which concerns the delay between two consecutive
copies above, should optimally be a random variable. In principle,
the delay should be longer than the burst channel fading period
length. But it cannot be too long, because too long will increase
the transmitting time of a data packet and decrease SR-ARQ
performance. Nevertheless, for a time-vary channel it is a very
difficult issue to define random burst channel fading period
length.
Referring to FIG. 4, a method of an interleaving procedure
according to the invention is shown, which is used to solve a
problem of inserting delay between every copy of the same lost data
packet. Before interleaving, there are 3 copies, &1, &1,
&1, of #1 data packet, 2 copies, &2 &2, of #2 data
packet and 2 copies, &3 &3, of #3 data packet queuing in
sequence in the queue. There is no delay between copies of the same
data packet, the sending principle is first-in-first-out. After
interleaving, between every & copy of the same # number data
packet, one copy of two other # number data packets is inserted.
For example, between every two copies of &1, a copy &2 and
a copy &3, two copies in total, are inserted, between two
copies of &2, a copy &3 and a copy &1, two copies in
total, are inserted, between every two copies of &3, a copy
&1 and a copy &2, two copies in total, are inserted. It is
formed that the delay time is 3, between every copy of same data
packet.
In principle the interleaving method of the present invention is
similar to interleaving in channel coding, but in the present
invention the interleaving object is data packets and not bits,
furthermore only the multiple copies of the retransmitted lost data
packet are interleaved, and the interleaving is before
retransmission. It is interleaving transmission. FIG. 4 shows after
interleaving, multiple copies, belonging to the same lost data
packet, are transmitted in sequence after delay. The delay time is
the time length of the number of different sequence number data
packets queuing in the interleaving queue. If the burst channel
fading period length is long, then the number of different sequence
number data packets queuing in the interleaving queue is more and
the delay is longer.
FIG. 5 shows an embodiment for the scheme of interleaving
retransmission in the invention. The transmitter needs to set three
queues, including transmission queue, retransmission queue and
buffer queue. The transmission queue, marked with #, is used to
store new transmitting data packets. The retransmission queue,
marked with &, is used to store multiple copies of specific
data packets needed to be retransmitted. The buffer queue, which
does not belong to the scheme of the invention, is used to store
the data packets having been transmitted but without receiving the
acknowledgement signal. Suppose that #6, #7 and #8 are arranged in
sequence from queue head in the transmission queue. In
retransmission queue there are arranged in sequence with two
copies, &1, &1, of #1 data packet, two copies, &2,
&2, of #2 data packet, and three copies, &3, &3,
&3, of #3 data packet. It is known from the retransmission
queue that the number of different retransmitted packets is 3. In
the retransmission queue, the number of the same data packet copies
is acquired, first according to the current retransmission number
then calculating with the linear increase principle or exponential
increase principle, all the copies of same data packet are stored
continuously.
In step 1, at first it is necessary to determine whether the
retransmission queue is empty or not; if the retransmission queue
is empty, every data packet in the transmission queue is
transmitted according to the first-in-first-out principle; if the
retransmission queue is not empty, then it enters the interleaving
transmission state.
Entering the interleaving transmission state is implemented by
setting the length of the interleaving retransmission interval. In
order to control the minimum value of multiple copies transmission
delay for specific data packet, the minimum value d.sub.s of
interleaving retransmission interval length should be set (it can
be determined by the specific mobile communication system). In the
figure, two conditions are set, they are d.sub.s=5 and d.sub.s=1,
respectively. The length of the interleaving retransmission
interval is a value which should be chosen as the largest one among
the minimum value (d.sub.s) of the interleaving retransmission
interval length and the maximum value of number (in the figure
example it is 3) of different data packets in retransmission
queue.
In step 2, interleaving transmission is proceeded, it is started to
form every interleaving retransmission interval.
When d.sub.s=5, the timer is set to 5. From the retransmission
queue, one copy of every retransmission data packet is selected,
&1, &2 and &3, and they are transmitted by the
first-in-first-out principle. When the transmission is finished,
and if the timer is not over, then data packets #6 and #7, in the
transmission queue, are transmitted by the first-in-first-out
principle, until the timer is over, the first interleaving
retransmission interval is ended. At this moment, if the
retransmission queue is not empty, the next interleaving
retransmission interval is started. Again, from the retransmission
queue, one copy of every retransmission data packet is selected,
&1, &2 and &3, and they are transmitted by the
first-in-first-out principle. When the transmission is finished,
and if the timer is not over, continuously data packet #8, in the
transmission queue, is transmitted; when it is finished, and if the
timer is not over, stop transmission until the timer is over, the
second interleaving retransmission interval is ended. At this
moment, if the retransmission queue is still not empty, the next
interleaving retransmission interval is started. Again, from the
retransmission queue one copy of #3 data packet, &3, is
selected and transmitted; when it is finished, and if the timer is
not over and the transmission queue is empty, stop transmission
until the timer is over, the third interleaving retransmission
interval is ended. As the retransmission queue is now empty, the
interleaving transmission will be ended, with first-in-first-out
principle, new data packet transmission in transmission queue is
started (if there are new data packets in the transmission
queue).
When d.sub.s=1, the timer is set to 1. From the retransmission
queue, one copy of every retransmission data packet is selected,
&1, &2 and &3, and they are transmitted by the
first-in-first-out principle; When the transmission is ended the
timer is over, so the first interleaving retransmission interval is
ended. Because the retransmission queue is not empty, the next
interleaving retransmission interval is started. Again, from the
retransmission queue, one copy of every retransmission data packet
is selected, &1, &2 and &3, and they are transmitted by
the first-in-first-out principle; when the transmission is
finished, the timer is over, so the second interleaving
retransmission interval is ended. At this moment, the
retransmission is still not empty, so the next interleaving
retransmission interval is started. Again, from the retransmission
queue one copy of #3 data packet, &3, is selected and
transmitted; when it is finished, the timer is over, so the third
interleaving retransmission interval is ended.
In step 3, when the retransmission queue is empty, the interleaving
retransmission state is ended; with the first-in-first-out
principle, data packets in the transmission queue are transmitted.
As shown in the figure when d.sub.s=1, after the third interleaving
retransmission interval is ended, it is started to transmit data
packets in transmission queue, #6, #7, and #8.
Two real examples in FIG. 5 show that the length of the
interleaving retransmission interval is the retransmission delay of
multiple copies of every lost packet. In reality, the length of the
interleaving retransmission interval is determined by d, and the
number of lost data packets; d.sub.s=1 is a special case, the
length of interleaving interval is only determined by the number of
successively lost packets.
With reference to FIG. 7, in reality, the interleaving
retransmission procedure of the invention is using the number of
continuously lost packets in the last transmission to predicate the
length of current channel fading period. In an ideal case, the
length of channel fading period is unchanged, only one time of
interleaving retransmission is needed, i.e. every lost data package
needs to transmit only two copies.
FIG. 7 shows that when the number of continuously lost packets in
the last transmission is 4 (#3, #4, #5, #6), the length of
predicting channel fading period is 4. After one time of
interleaving retransmission, that is every lost data packet only
transmits two copies, &3, &4, &5, &6 and &3,
&4, &5, &6, the receiver receives the retransmission
data packets &3, &4, &5, &6 successfully.
Therefore, the value of the timer is decided, based on the number
of continuously lost packets in the last transmission or based on
the measurement result of the technique by which the fading pitch
is measured.
Referring to FIG. 6, it shows another embodiment of the method for
multiple copies plus delay retransmission scheme of the invention.
It is a multiple queues polling retransmission method. Compared
with the embodiment shown in FIG. 5, the interleaving transmission
method shown in FIG. 6 is different.
The transmitter also needs three queues, including transmission
queue, retransmission queue and buffer queue. The transmission
queue is used to store new data packets to be transmitted. The
retransmission queue is used to store multiple copies of every
specific data packet needed to be retransmitted. The buffer queue
is used to store the data packets having been transmitted but
without receiving the acknowledgement signal.
When the retransmission queue is not empty, it enters an
interleaving retransmission state, polling transmission is
started.
Step 1. Set N individual copy queues, the sequence numbers are copy
queue 1, copy queue 2 , copy queue 3, . . . , copy queue i, . . . ,
copy queue N, one copy of every specific retransmission packet will
be stored in each copy queue in sequence. For example, two copies
of #1 retransmission data packet, &1 and &1, are stored in
copy queue 1 and copy queue 2 , respectively, two copies of #2
retransmission data packet, &2 and &2, are stored after
copy &1, in copy queue 1 and copy queue 2 respectively; three
copies of #3 retransmission data packet, &3, &3 and &3,
are store, after copy &2, in copy queue 1, copy queue 2 and
copy queue 3 respectively. This means if a retransmission data
packet has k individual copies, then the k copies are stored in
copy queue 1, copy queue 2 , . . . , copy queue k,
respectively.
Step 2. Starting from copy queue 1 to copy queue N, they are
transmitted in sequence. Then, if any copy queue is not empty, it
will start again from copy queue 1 to copy queue N transmitted in
sequence until all the copy queues are empty, and the polling
transmission procedure is ended. When transmitting, only after the
current copy queue is empty, then the next copy queue can be
transmitted. This means only when the i.sup.th copy queue is empty,
then the (i+1).sup.th," copy queue can be transmitted. Furthermore
in spite of whether there is any empty copy queue from i.sup.th
copy queue to N.sup.th copy queue in real, it must be transmitted
in sequence until the N.sup.th copy queue.
Step 3. When all N individual copy queues are empty, the polling
transmission procedure is ended, and transmitting new data packet
in the transmission queue is started. Considering that in many
mobile communication systems, the numbers of retransmission are
controlled in real operation, for example, in IS-99, the numbers of
retransmission cannot exceed 3. This means that in one
retransmission, the copy number will not exceed a certain fixed
value. If using the linear increase method of the invention, then
in one retransmission the copy number does not exceeded 4. If using
the exponential increase method of the invention, then in one
retransmission the copy number does not exceeded 8. When using the
polling transmission of multiple queues, the number of copy queue
can be set respectively as 4 and 8, and the implementing procedure
will be greatly simplified.
Therefore, when limiting the numbers of retransmission, the effect
of the second method, polling retransmission method, is same as the
first method with d.sub.s=1, but the implementation of polling
retransmission method is simpler.
Reference is now made to FIG. 8, FIG. 9 and FIG. 10. Under the
conditions of a 20 db signal-to-noise ratio (S/N), and different
channel fading speed f.sub.d (Hz), such as 5 Hz (represented by
triangle), 10 Hz (represented by lozenge) and 50 Hz (represented by
*), with an emulation approach, shown are the result of the
relationship between delay time (T.sup.delay) and effective
throughput, delay time (T.sub.delay) and the mean of the number of
retransmissions, and delay time (T.sub.delay) and variance of the
number of retransmissions, respectively. In the figures, a solid
line represents Delay-Linear-Multiple-Copies-Retransmission scheme,
dot dash line represents
Delay-Exponent-Multiple-Copies-Retransmission scheme.
The figures show that slow fading speed will lead to decreased
throughput, increase mean of the number of retransmission and
variance of the number of retransmission. This is because slow
fading speed has a larger correlation and longer fading period
length, which increases the probability of successive lost
packets.
Besides, directing to the traditional Selective Retransmission (SR)
scheme, the Delay-Linear-Multiple-Copies-Retransmission
(D-Linear-MCR) scheme and the Delay-Exponent-
Multiple-Copies-Retransmission (D-Expo-MCR) scheme, it can further
be seen the influence of channel round trip time to effective
throughput, mean of number of retransmissions and variance of
number of retransmissions. The result is that: along with increased
round trip time, the Delay-Multiple-Copies-Retransmission scheme of
the invention will get better performance. This is because after
passing a longer round trip time, the correlation of successive
retransmission is decreased.
In addition, also directing to the traditional Selective
Retransmission scheme, the
Delay-Linear-Multiple-Copies-Retransmission scheme and the
Delay-Exponent-Multiple-Copies-Retransmission scheme, with
different channel fading speed (as 10 or 100 Hz), it can also be
seen the influence of signal-to-noise-ratio (SNR-db) to effective
throughput, mean of number of retransmissions and variance of
number of retransmissions.
The result is: along with the decrease of signal-to-noise-ratio,
especially when the signal-to-noise-ratio is less than 25 db, the
performance of delay multiple copies retransmission is decreased
greatly.
FIG. 11 is a simulation curve diagram of transmission performance
improvement according to the method of the invention. In a typical
mobile communication system, emulating different ARQ schemes, the
performance of Transfer Control Protocol (TCP) have been acquired.
In emulation, the following conditions are taken: The capacity of
wired network is 100 Mbps; The capacity of wireless link is 2 Mbps;
The propagation delay of wired network is 50 ms; The propagation
delay of wireless network is 10 ms; The link layer packet size is
53 bytes; The TCP packet length is 576 bytes; The buffer in the
base station can store at most 384 link layer packets; The maximum
times of retransmission are 2 (copies are 3 or 4); The minimum
value of interleaving retransmission interval d.sub.s is 1; Two
state Markovian model is used to simulate fading channel.
FIG. 11 compares the end-to-end TCP throughput of different ARQ
schemes. The X-axis represents channel fading speed f.sub.d (Hz),
and the Y-axis represents throughput. The dot line connected with
hollow lozenge blocks, shows the ideal state without limiting
numbers of retransmission (unlimited retransmission). The solid
line connected with solid square blocks, shows performance of
traditional selective retransmission (SR-ARQ) scheme. The solid
line connected with solid lozenge blocks, shows performance of
Linear-Multiple-Copies-Retransmission (Linear-MCR) scheme. The
solid line connected with + symbol, shows performance of
Exponent-Multiple-Copies-Retransmission (Expo-MCR) scheme. The
solid line connected with solid triangle shape, shows performance
of Delay-Linear-Multiple-Copies-Ret (D-Linear-MCR) scheme. The
solid line connected with + shape, shows performance of
Delay-Exponent-Multiple-Copies-Retransmission (D-Expro-MCR) scheme.
The results show: the invention scheme compares with traditional
SR-ARQ scheme, a better TCP end-to-end throughput can be acquired.
Experience also shows performance of the D-MCR scheme is better
than MCR scheme, while different multiple copies schemes (Linear
and ExPro) have no much difference.
In general, in the fast fading channel, because the mean lost rate
of data packets is higher, performance of the Exponent scheme is
better than the Linear scheme, throughput is large. But in the slow
fading channel, longer fading period length will greatly decrease
the effectiveness of the multiple copies scheme, so at the lower
effective throughput condition, throughput of the Exponent scheme
will be lower than the Linear scheme. Therefore, it should select
different multiple copies scheme for different requirement.
* * * * *